Patients’ perspectives on prescription cannabinoid therapies: a cross-sectional, exploratory, anonymous, one-time web-based survey among German patients

2.2. Setting

The original intent of this study was to draw a representative sample of patients treated with cannabinoids in the two German federal states of Berlin (city state) and Lower Saxony. SARS-CoV-2 pandemic and its contact regulations reduced the number of physical visits of patients to their doctor’s practices and lowered thus the recruitment rate. To counter that, the state of Brandenburg was added in an amendment on January 20, 2022. Physicians with specialist qualifications in anesthesiology, general and family medicine, neurology, and internal medicine, who hold a statutory health insurance contract in one of the federal states, were initially contacted in a stratified procedure, as well as outpatient tertiary centers. These specialist groups were selected because, according to the aforementioned surveillance survey, they account for 88.6% of cannabinoid prescriptions in Germany (6). Physicians were contacted by e-mail or fax to invite their patients treated with cannabinoids to participate in this anonymous web-based cross-sectional survey. In a second step, the same physicians were also contacted by telephone, in random order and in quotas that reflected the proportion of cannabinoid prescriptions for each specialty group, because we did not expect a sufficient response in the first step. Calls were to be made until the target number of participants was reached. Due to the late amendment toward the end of the recruitment phase, this second step could not be implemented in the state of Brandenburg.

2.3. Participants

As is common for exploratory studies, the sample size was not calculated based on previously described effect sizes but set to n = 300 owing to resources. The target sample size of n = 300, however, is sufficient to detect (descriptive) effects with an effect size of Cohen’s d > 0.17 and thus to distinguish large (d > 0.80), moderate (d between 0.50 and 0.80), small (d between 0.20 and 0.50) and negligible effects (d < 0.20) from one another (14).

Interested physicians received informational materials and survey access codes to distribute to their cannabinoid patients, including those who had completed or discontinued such therapies in the previous 12 months. With information and access code, patients could later decide at home whether or not to participate and provide all information discreetly and anonymously. The cooperating physicians were compensated for their time and effort in inviting patients to the study with 50 Euros per patient who eventually participated (13). Each unique participation login code could only be used once to complete the web-based survey. Thus, the study center did not receive personalized or pseudonymized participant data, and the prescribing physicians did not have access to study data (Figure 1). Active recruitment ended on May 31, 2022; the last participant completed this survey on June 12, 2022. More details can be found in the protocol publication (13).

2.4. Outcome parameters and variables

In this survey, subjects were asked to complete all questionnaires and questions regarding their cannabinoid therapy twice in the same session: once for the current time point (i.e., at the date of the survey participation), and then a second time for the time point prior to the commencement of the cannabinoid therapy, by self-recalling that period (cf. Figure 2A) (13).

All participants completed the Treatment Satisfaction Questionnaire for Medication (TSQM, version II) (15, 16), the WHO-5 quality of life questionnaire (17), the Patient Reported Outcome Measurement Information Systems (PROMIS-29), numeric rating scales (NRS) for the daily average, lowest and strongest pain (18, 19), and the Faces Pain Scale (20, 21). In addition, the following validated questionnaires were displayed to the patients suffering from the respective conditions: the Hospital Anxiety and Depression Scale (HADS) (22–25) for patients with symptoms of depression or anxiety, the Multiple Sclerosis Spasticity Scale 88 (MSSS-88) (26) for participants with multiple sclerosis, the Anorexia Nervosa Inventory for Self-Rating (ANIS) (27) and the Minimal Documentation System on Distressing Symptoms (MIDOS2) for participants with anorexia-related and non-anorexia-mediated weight loss and nausea/vomiting (28), respectively, and the Adult ADHD Self-Report Scale (ASRS) (29, 30) for participants with ADHD-associated symptoms. Licenses were obtained for questionnaires prior to the commencement of the study, wherever required.

Sociodemographic data were collected (age, occupational status, highest level of education, relevant main diagnoses, previous and current other therapies, medication, side effects) including previous experiences with pre-existing or symptom-related psychological trauma, and relaxation techniques or psychotherapy, patients’ personal opinions about cannabinoids, and varying experiences with different cannabis-based medications. These included the exact cannabis preparation, dose, mode of application, adverse effects, and reasons for pre-prescription use of cannabinoids (if applicable). The duration of cannabis therapy was asked, as well as the symptoms for which it was prescribed. The most common symptoms (6) of pain, spasticity, weight loss, nausea/vomiting, depression, attention-deficit/hyperactivity disorder (ADHD), and a free-text field were available as possible answers. Underlying conditions were asked as free text and ICD number.

Participants were also asked about details of the cannabinoid prescription process, particularly if there were any problems in obtaining the cannabinoid medication. Response options ranged from “very major problems” (+2) to “no problems at all” (−2). If problems were reported, participants were asked at which point in the health care system those had occurred (family doctor, specialist, health insurance company, pharmacy, elsewhere [free text]). Health insurance status was queried (as between (i) statutory health insurance with or without specialized supplementary private insurances, e.g., for certain dental treatment or eyewear, (ii) complete private health insurance, (iii) none at all or (iv) something else [free text]. For details about German health insurance see reference (31)). General attitudes toward cannabinoids, before and after the start of cannabinoid therapy, were rated on a 5-point Likert scale, ranging from very positive (+2) to very negative (−2). It was asked what these attitudes were based on (e.g. experiences, media reports, personal experiences, free text). Changes in cannabinoid medication were asked, as were the reasons for any changes (insufficient effect, side effects, interactions, therapy no longer necessary, reason not known, free text). In addition, it was asked from whom the idea for cannabinoid therapy originally came (family, friends/acquaintances, treating physician, patient him/herself, free text). (For details, see reference 13).

2.5. Data sources and measurement

The survey (Supplement 1) was implemented using Lime Survey software (Lime Survey GmbH, Hamburg, Germany, v 5.4). Informed consent was declared anonymously online and was a prerequisite for data entry along with the login code. Software and data were stored on a server hosted and secured by Charité – Universitätsmedizin Berlin. The study was approved by the Charité ethics committee on December 16, 2020 (EA1/327/20) and registered at the German Register of Clinical Studies (DRKS00023344).

3.2. Participants

A total of 486 invitations with individual login codes were sent to these physicians (Figure 3A), of which 256 were activated by the participants to start the survey. 132 (of 257) by patients of anesthesiologists, 45 (of 107) by patients of general practitioners, 28 (of 42) by patients of neurologists, and 49 (of 80) by patients of internists. Of the 256 surveys started, 40 were discontinued after the first few questions and were therefore excluded from the analysis (because the study was conducted anonymously, we do not have details about their reasons for discontinuing the survey. However, we speculate that they mistakenly expected to be able to resume the survey later.) The remainder, a total of n = 216 participants, 112 male and 104 female, completed the survey.

The originally planned number of n = 300 participants could not be reached until the end of May 2022, despite intensive recruitment efforts and an extension of the recruitment period. When we contacted physician practices, we repeatedly received feedback that there was no capacity to support the study due to the pandemic. We believe that significantly more physicians would have invited their cannabinoid patients to participate under pre-pandemic conditions. Due to limited human resources and study budget, the study was stopped at n = 216 completed surveys included in the final data analysis. However, the reached participant number of n = 216 is still sufficient to detect effects with effect sizes of Cohen’s d > 0.195, so that the distinction between the main effect sizes (small, moderate, large) is still ensured (14).

The mean age of cannabinoid patients was 51.8 (± 14.0) years (for details see Table 1). Participants reported to have been on prescription cannabinoid therapy on average for 2.4 ± 1.4 years. The highest educational attainment reported by n = 50 (23%) was high school graduation, n = 58 (27%) was an apprenticeship, n = 34 (16%) was a college degree, n = 56 (26%) was a university degree. N = 82 (38%) of the participants were working, n = 85 (39%) retired, and n = 37 (17%) permanently ill (other responses: n = 12, 6%).

3.3. Diagnoses

The most common symptoms based on which prescription was made were chronic pain (n = 155, 72%), spasticity (n = 17, 8%), depression (n = 8, 4%), ADHD (n = 4, 2%) (remainder: n = 32, 15%). This is in-line with numbers reported in a previous German study (6). Additionally, participants were asked to enter all diagnoses known to them from their medical records in free text and, if available, along with the corresponding ICD-10 codes (32). Missing ICD codes were post-coded by study staff based on participants’ free text entries and validated by the study physician. Participants were also asked which of these diagnoses were primarily to be alleviated by cannabinoid therapy. In most cases one (n = 89, 41%), two (n = 45, 21%) or three (n = 24, 11%) different diagnoses were given (see Figure 3B). The most frequent diagnoses by International Classification of Diseases 10th revision (ICD-10) were (Figure 3C) chronic pain (R52), somatoform disorders (F45), other soft tissue disorders, not elsewhere classified (M79), and other intervertebral disc disorders, incl. Thoracic, thoracolumbar and lumbosacral disc disorders (M51) (cf. Figure 3C).

3.4. Attitudes toward cannabis and cannabinoids

Before presenting “pre-post” comparisons (current health status compared to self-reported health status before cannabinoid therapy in retrospect) below, we explicitly point out to the fact that these were calculated from a one-time cross-sectional survey (cf. Figure 2A). The presented exploratory significance levels and effect sizes must therefore be interpreted with appropriate caution and due restraint and no causal attributions are being derived from this exploratory observational data set (cf. the Discussion section).

Most patients reported to have had a neutral or mildly positive opinion of cannabinoids before starting cannabinoid therapy (Figure 2B). This was based in roughly equal measure on their own experiences with cannabis, on statements by their friends and family, and on media coverage. After the start of their cannabinoid treatment, most participants (n = 204, 94%) reported a positive attitude toward cannabinoids. Most of them based this on their own experiences and those of their friends. The shift toward more positive attitudes (on a 5-point Likert scale from −2 to +2) was even greater among participants without prior experience (+0.33 ± 1.11 to +1.61 ± 0.65) than for patients with prior experience (+0.81 ± 0.91 to +1.73 ± 0.51).

Previous experience with cannabis prior to medically prescribed cannabinoids was reported by n = 93 (43%) of participants, of which n = 64 (30%) reported having had taken it for medical reasons, n = 38 (18%) for recreational purposes, n = 28 (13%) out of curiosity, n = 35 (17%) in peer-group settings, n = 2 (1%) reported having tried cannabis due to peer pressure and none for religious reasons (multiple entries possible).

3.5. Experiences regarding the prescription process

Participants reported that the current cannabinoid prescription process had been suggested first by the participants themselves in n = 101 (47%) of cases, in n = 86 (40%) by the treating physician, and in n = 11 (5%) and n = 8 (4%) by friends and family members, respectively. The majority of participants (n = 175; 81%) had statutory health insurance coverage, and n = 34 (16%) were privately insured, which roughly corresponds to the average in the German population (31, 33). Of those holding private insurance, n = 31 out of 34 (91%) were reimbursed by their private health care insurer for the cost of cannabinoid prescriptions. In contrast, only 133 out of 175 (76%) of those with statutory health insurance were reimbursed by their health insurer. Accordingly, the remaining 42 of 175 (24%) paid for their cannabinoid medications out of pocket.

N = 70 (31%) of participants reported encountering barriers to accessing cannabinoid treatment (n = 39, 18% minor problems and n = 31, 13% major problems). The problems were primarily with the health insurance company (n = 56; 26%). N = 24 (11%) were with the specialist, n = 17 (8%) with the primary care physician, and n = 8 (4%) with the pharmacy. N = 146 (69%) of the participants indicated that they did not experience any problems and found it easy or very easy to get prescription cannabinoids.

3.6. Cannabinoid preparations and changes in therapy

Cannabinoids were prescribed in the form of cannabis flowers (n = 97; 45%), cannabis extracts (n = 52; 24%), dronabinol (equivalent to marinol; n = 43; 20%), and nabiximols (n = 27; 13%; multiple answers permitted). Of the n = 149 participants with a prescription for cannabis flower or extracts, n = 16 (11%) took it frequently or always with tobacco, n = 13 (7%) occasionally.

On average, participants had received prescription cannabinoids for 1–2 years at the time of the survey. During their therapy, most of the participants’ cannabinoid therapy was modified at least once (n = 139 out of n = 216, i.e., 64%). In those n = 139 participants the dose was either increased (n = 83, 60%) or reduced (n = 21, 15%), the medical cannabis cultivar was changed (n = 57, 41%), the mode of application, e.g., from drinking tea to vaporization (n = 9, 6%), and/or the medicinal product was changed, e.g., from nabiximols to dronabinol (n = 35, 25%).

The reason for these changes were in n = 100/139 cases (72%) a desired change in the effectiveness of the drug, in n = 27 (19%) occurrence of adverse side effects and in n = 5 (4%) interactions with other drugs. Among patient-reported adverse events, the most commonly reported were dry mouth (n = 14, 7%), attention-deficit (n = 11; 5%), fatigue (n = 10, 5%), dizziness (n = 10, 5%), and somnolence (n = 9, 4%).

3.7. Questionnaires: perceived changes in symptoms and well-being

The results for the validated questionnaires on how participants’ physical and mental symptoms and their perceived quality of life subjectively changed during cannabinoid therapy are shown in Table 2.

Satisfaction with the treatment (measured by the TSQM II, Figure 2C) increased in the three subscales perceived effectiveness, side effects and overall satisfaction, while the ease of taking medication (“convenience”) rose (all with large effect sizes and p < 0.001).

In accordance with this, well-being (as measured by the WHO-5, Figure 2D) increased from 6.0 ± 4.6 to 13.8 ± 5.8 out of 25 points (d > 1.0; p < 0.001).

In the PROMIS (cf. Figure 2E and Table 2) participants reported having fewer problems with their bodily functions (physical functioning). They experienced less anxious and depressive feelings and their fatigue decreased as well. Similarly, they reported fewer problems fulfilling their social roles and the pain was perceived to have less impact on their daily lives (pain interference scale). The only exception to the large perceived effects found in this questionnaire was sleep disturbance, which showed only a small improvement over time.

N = 155 (72%) participants indicated that pain relief was the primary reason for their cannabinoid prescriptions. These were asked in greater detail about their perceived pain intensities. On a 6-point face pain scale average pain intensity decreased from 2.9 ± 1.4 to 1.0 ± 1.2 (d > 1.0, p < 0.001). In addition, pain was rated on an 11-point numerical rating scale from 0 to 10 (NRS) (Figure 2F) for the most severe, average, and least severe pain intensity during the day, and had decreased in all three (d > 1.0, p < 0.001). Finally, the pain reduction experienced through the daily medication (on a scale of 0 to 100) increased from 37.4 ± 24.7 to 59.8 ± 23.6 (d = 0.93, p < 0.001).

Eight participants who reported anxiety or depression as their primary health problem completed the HADS questionnaire described substantial reductions of both anxiety (HADS-A) and depressive moods (HADS-D), thus changing from classified as “severe (case)” (scores >10.5) to “moderate (borderline case)” (scores between 7.5 and 10.5) or even “inconspicuous (non-case)” (scores ≤7.5).

In n = 17 participants with multiple sclerosis, the MSSS-88 was used. Self-ratings for muscle stiffness decreased with low effect size, whereas scores for spasticity, impairment of walking ability, and impairment of body mobility decreased with moderate to large effect sizes (p < 0.001).

Thirty-two patients with weight loss without anorectic symptoms completed the MIDOS questionnaire. The global symptoms’ sum scale (ranging from 0 to 30) decreased from 15.4 ± 7.8 to 5.9 ± 3.9 (d > 1.0, p < 0.001), indicating clinically significant amelioration of symptoms.

Since only n = 4 participants with ADHD symptoms participated in the study, their results in the ASRS are not presented here to avoid over-interpretation.

None of the participants filled out the ANIS questionnaire for anorexia nervosa.

Special thanks to Nadine Sylvester, Miriam Rösner, Gunda Loibl, Carsten Schütz, and Arnab Chakrabarty, without whom this study would not be possible, and to Daniel Itzlroither and Manfred Wischnewsky for their support.